One of the classics I run into when I review how a business measures itself: a dashboard with 30 or 40 indicators, charts for everything, and nobody in the meeting able to say which of those numbers defines whether the month was good or bad. That is why it is worth going back to basics and explaining what a KPI is, because in most companies there are plenty of numbers and very few indicators anyone actually uses to make a decision.
The acronym stands for Key Performance Indicator. And even though it sounds like consulting jargon, the underlying idea is fairly simple: out of all the numbers your business generates, KPIs are the few that tell you whether you are getting closer to your goals or drifting away from them.
What is a KPI, exactly?
A KPI is a metric that meets two conditions: it measures progress toward a concrete business objective, and someone reviews it at a defined frequency in order to decide what to do.
The analogy I like best is a car's dashboard. A modern car generates hundreds of data points per second (temperature of each cylinder, pressure of each tire, revs, voltages), and the onboard computer uses all of them, but what the dashboard shows you is three or four things: speed, fuel, temperature. Those are the ones that change how you drive right now. A KPI is exactly that, the number that earned a spot on the dashboard. The rest of your metrics can live happily in an appendix spreadsheet, available whenever something needs diagnosing.
And what is it for, concretely? Three things: knowing whether you are on track against a goal, catching problems before they get expensive, and aligning a team around what matters. When a team shares two or three clear KPIs, the tone of the discussions changes on its own, because people stop arguing about what to look at and start arguing about what to do with what everyone is looking at.
How is a KPI different from a metric?
Every KPI is a metric, but very few metrics deserve to be a KPI. A metric is anything you can measure: visits, followers, calls, units shipped. It becomes a KPI when it gets tied to an objective and a decision.
The important nuance is that the same metric can be a KPI in one business and filler in another. For a consumer brand that lives off awareness, social media reach can be a central indicator. For a B2B company that closes three large contracts a year, that same reach is a nice-to-know and little more: their KPIs will live in meetings booked and proposals sent, which are the early stages of their sales funnel.
In fact, two metrics I have already covered in detail on this blog, CTR and engagement, are good examples: almost always useful as diagnostics, KPIs only when the business objective actually runs through them.
What makes a good KPI?
After years of looking at other people's dashboards, I landed on a short list of four tests.
First, it is connected to a written objective. If you cannot complete the sentence "this number measures our progress toward X", you have one more metric, not a key indicator.
Second, it has an owner. A person with a first and last name answers for that number, reviews it and proposes what to do when it moves. When the owner is "the team", in practice nobody looks at it.
Third, it drives decisions. The test I use most: if this number goes up or down 20% next month, what would we do differently? If the answer is "nothing, we would carry on as usual", that number is not doing the job of a KPI, however nice it looks on a chart.
Fourth, it has a target and a review frequency. A number on its own says very little: a 4% conversion rate can be excellent or worrying depending on your history and your category (the percentage is made up, for illustration; the context is something you build with your own data series).
There is one extra detail that separates teams that measure well: they combine lagging and leading indicators. Lagging ones confirm results that already happened, like the quarter's closed sales. Leading ones move earlier and give you time to react, like meetings booked or quotes sent this month. If all your KPIs are lagging, you find out about problems when they are already history and the only thing left to do is explain them in the meeting.
Which KPIs should you use? Examples by area
The examples below are illustrative, meant to ground the idea; yours depend on your objectives, not on a generic list.
In sales: close rate (proposals won over proposals sent), sales cycle length and average ticket. In finance: gross margin, cash flow and average days to collect, which is the indicator that prevents the most scares in small companies. In operations: delivery time and percentage of orders with errors. In customer service: first response time and cases resolved on first contact.
And in digital marketing, which is where mile-long dashboards get abused the most, the sensible KPIs tend to be few: acquisition cost, conversion rate by channel and, for the organic channel, things like the conversions it originates and searches for your brand, well beyond plain sessions. On the metrics for measuring organic seriously, including visibility in AI answers, which is already part of the job, I wrote a full guide in the metrics CMOs and Growth Managers should be using.
How do you choose the KPIs for your business?
Here is the full recipe, step by step.
First, write down the objective before opening any tool. "Grow revenue in segment X", "lower the cost to serve", "reduce customer churn". Without a written objective, every metric looks important and the dashboard grows on its own.
Second, build the list of candidates and run them through the decision test: what would we do differently if this number moved sharply? The ones with no answer go to the appendix spreadsheet, guilt-free. They still exist there in case something odd needs diagnosing.
Third, keep only a few. Between three and five per team is a healthy range. With ten, the results meeting turns into a tour of charts where none of them gets properly discussed.
Fourth, document each chosen KPI: exact formula, data source, owner, review frequency and target. The exact formula matters more than it seems. I have watched very long arguments between two areas reporting "conversion" with different definitions, and both were right according to their own spreadsheet.
Fifth, pair one lagging and one leading indicator per objective, so you get the result and the early signal at the same time.
Sixth, schedule a quarterly review of the KPIs themselves. Objectives change, and an indicator that was central a year ago may be obsolete today. Deleting old KPIs is as healthy as defining new ones, even if letting go of the chart stings a little.
Which mistakes should you avoid?
The most common one is filling up on vanity metrics: numbers that only ever go up and are therefore a pleasure to report, like cumulative followers or total historical visits. They keep climbing even when the business is flat, so they tell you nothing about your objectives. On how to separate data that helps from data that just decorates, I wrote more in data-driven marketing.
The second is measuring what is easy instead of what is important. The tool hands you sessions and clicks instantly, while calculating real acquisition cost requires crossing sources and getting areas to agree, so guess which of the two ends up filling the dashboard.
The third is copying the KPIs of another company or of some benchmark from abroad. That company has objectives, margins and markets different from yours, and its indicators come tied to those conditions, not to yours.
And the fourth is changing a formula midway without leaving a record. If in March "active customer" meant one thing and in August it means another, your historical series is now useless and year-over-year comparisons become well-intentioned fiction.
This week's takeaway
Open the dashboard you use today and ask every number the same question: if this goes up or down 20% next month, what would we do differently? The ones with a clear answer are your KPIs. Write down an owner, a target and a review frequency for each, ideally in the same spreadsheet, so the definition is on record. Move the rest to an appendix tab, where they remain available without stealing attention. The exercise is free, takes an afternoon, and the next results meeting will show it right away: fewer charts, more conversation about what to do with the numbers that made the cut.